Sagnac Interferometer Microscope

• Sagnac interferometer microscope is an optical setup that uses counter-propagating beams to cancel reciprocal signals while isolating weak non-reciprocal effects.
• It delivers significant improvements in signal-to-noise ratio for applications such as particle tracking, magneto-optic Kerr imaging, and spin-caloritronic measurements.
• Adaptations include all-fiber and optical tweezers configurations that enhance angular resolution and spatial precision for advanced scientific research.

A Sagnac interferometer microscope is an optical measurement configuration in which two counter-propagating beams traverse substantially the same optical path in opposite directions and are recombined so that large reciprocal backgrounds are suppressed while non-reciprocal or antisymmetric signal terms remain measurable. In the cited literature, this architecture appears in counterpropagating optical tweezers for position tracking and in magneto-optic Kerr microscopes, including an all-fiber scanning implementation at 820 nm. Reported performance includes a more than 5 times improvement in the signal-to-background ratio, corresponding to a more than 30-fold improvement of the signal-to-noise ratio for particle tracking, and better than $1~\mu\mathrm{rad}$ angular resolution with $1.5~\mu\mathrm{m}$ spatial resolution for Kerr imaging (Galinskiy et al., 2015, Fried et al., 2014).

1. Operational principle

The Sagnac interferometer has historically been used for detecting non-reciprocal phenomena, such as rotation. In a zero-area, all-fiber loop, two time-reversed, linearly-polarized beams propagate clockwise and counterclockwise and are recombined so that purely reciprocal phase shifts cancel out. In magneto-optic operation at normal incidence, the relevant non-reciprocal quantity is the Polar Kerr effect, for which the circular components acquire different phases and one defines the Kerr rotation by

$\phi_R-\phi_L=2\theta_K,$

so that the Sagnac phase is

$$\Delta\phi_{\rm Sagnac}=\phi_{\rm CW}-\phi_{\rm CCW}=2\theta_K.$$

An equivalent form,

$$\Delta\phi_{\rm Sagnac}=\frac{4\pi nL}{\lambda}\theta_K,$$

emphasizes that the same $\theta_K$ tags the two beams twice and that a long loop gives more optical gain (Fried et al., 2014).

For particle tracking, the same interferometric logic is expressed in terms of parity. Let $\phi_0(x)$ be the symmetric input beam amplitude and $F[x_p]$ the forward-scattering operator of a particle displaced by $x_p$. The scattered field is decomposed as

$$F[x_p]\phi_0=\phi_s(x)+\phi_a(x), \qquad P\phi_s=+\phi_s,\quad P\phi_a=-\phi_a,$$

where $1.5~\mu\mathrm{m}$0. After splitting, propagation, scattering, and recombination, the dark-port field is

$1.5~\mu\mathrm{m}$1

This form shows that the symmetric trapping background is weighted by $1.5~\mu\mathrm{m}$2, whereas the antisymmetric position-sensitive component is weighted by $1.5~\mu\mathrm{m}$3. For small oscillations, $1.5~\mu\mathrm{m}$4, the signal-to-background ratio becomes

$1.5~\mu\mathrm{m}$5

while the traditional back-focal-plane limit is

$1.5~\mu\mathrm{m}$6

The SBR gain is therefore

$1.5~\mu\mathrm{m}$7

and, since shot-noise-limited SNR scales as $1.5~\mu\mathrm{m}$8, the SNR gain is

$1.5~\mu\mathrm{m}$9

This suggests that Sagnac microscopy is effective when the reciprocal carrier can be attenuated more strongly than the non-reciprocal or antisymmetric signal term (Galinskiy et al., 2015).

2. Instrument architectures

In the counterpropagating optical-tweezers implementation of Galinskiy et al., a DPSS $\phi_R-\phi_L=2\theta_K,$0 laser source of up to $\phi_R-\phi_L=2\theta_K,$1 is split by PBS1+HW1 between a vertical levitator and the tweezers subsystem. A non-polarizing beamsplitter divides the tweezers beam into clockwise and counterclockwise arms of the Sagnac loop. Polarization is managed with HW2 and HW3, and a Brewster-tilted glass plate in the CCW arm reflects away excess power to equalize CW and CCW power at the trap. Two identical long-working-distance aspheres, LA1 and LA2, with $\phi_R-\phi_L=2\theta_K,$2, focus the beams to a common focal point and then recollimate them. After the trap, the beams recombine at the beamsplitter, interfere destructively toward the quadrant photodiode dark port and constructively back toward the laser. A polarizer after the dark port removes backscattered and reflected light, and lens L1 images the back-focal plane onto a First Sensor QP50-6 quadrant photodiode. A weak ghost reflection from the beamsplitter provides a traditional single-beam back-focal-plane signal for direct comparison (Galinskiy et al., 2015).

In the all-fiber Kerr microscope of Fried, Fejer, and Kapitulnik, the source is a superluminescent diode centered at $\phi_R-\phi_L=2\theta_K,$3 with $\phi_R-\phi_L=2\theta_K,$4 and coherence length $\phi_R-\phi_L=2\theta_K,$5. A polarization controller, two PM isolators, and a 3-port PM circulator provide at least $\phi_R-\phi_L=2\theta_K,$6 back-reflection isolation. A $\phi_R-\phi_L=2\theta_K,$7, $\phi_R-\phi_L=2\theta_K,$8 PM fiber beam-splitter launches the light into a $\phi_R-\phi_L=2\theta_K,$9 PANDA-type PM-fiber loop containing an inline LiNbO$$\Delta\phi_{\rm Sagnac}=\phi_{\rm CW}-\phi_{\rm CCW}=2\theta_K.$$0 waveguide EOM of approximately $$\Delta\phi_{\rm Sagnac}=\phi_{\rm CW}-\phi_{\rm CCW}=2\theta_K.$$1 length. All connector ends are $$\Delta\phi_{\rm Sagnac}=\phi_{\rm CW}-\phi_{\rm CCW}=2\theta_K.$$2-APC. In the free-space sample arm, a collimating aspheric lens with fiber-side $$\Delta\phi_{\rm Sagnac}=\phi_{\rm CW}-\phi_{\rm CCW}=2\theta_K.$$3, sample-side $$\Delta\phi_{\rm Sagnac}=\phi_{\rm CW}-\phi_{\rm CCW}=2\theta_K.$$4, and $$\Delta\phi_{\rm Sagnac}=\phi_{\rm CW}-\phi_{\rm CCW}=2\theta_K.$$5, together with a zero-order quarter-wave plate at $$\Delta\phi_{\rm Sagnac}=\phi_{\rm CW}-\phi_{\rm CCW}=2\theta_K.$$6, prepares the illumination and reflection path. Recombined light is routed by the circulator to a high-speed GaAs photodiode of $$\Delta\phi_{\rm Sagnac}=\phi_{\rm CW}-\phi_{\rm CCW}=2\theta_K.$$7 bandwidth, and two lock-ins read $$\Delta\phi_{\rm Sagnac}=\phi_{\rm CW}-\phi_{\rm CCW}=2\theta_K.$$8 and $$\Delta\phi_{\rm Sagnac}=\phi_{\rm CW}-\phi_{\rm CCW}=2\theta_K.$$9 (Fried et al., 2014).

The TSSE measurement discussed by Kimling and Kuschel implies a third architecture. A single, linearly polarized laser beam is split into two counter-propagating beams by a non-polarizing $$\Delta\phi_{\rm Sagnac}=\frac{4\pi nL}{\lambda}\theta_K,$$0 beam splitter. Each beam traverses the same optical elements in opposite directions: a quarter-wave plate, a focusing objective that forms a sub-$$\Delta\phi_{\rm Sagnac}=\frac{4\pi nL}{\lambda}\theta_K,$$1 spot, and the sample. Before recombination, a non-reciprocal phase modulator imposes a time-varying differential phase $$\Delta\phi_{\rm Sagnac}=\frac{4\pi nL}{\lambda}\theta_K,$$2. After interference at the original beam splitter, a photodiode detects the intensity and lock-in demodulation extracts the small magneto-optical contribution. The comment explicitly notes that neither it nor its references give a full parts list or specific mirror, lens, or detector models (Kimling et al., 2018).

3. Sagnac-enhanced optical tweezers microscopy

The particle-tracking lineage proceeds from the proposal by Taylor et al. to the experimental generalization by Galinskiy et al. Taylor et al. assumed Rayleigh scatterers trapped at an antinode in a standing wave, with identical forward and backward scattering and a single polarization. Galinskiy et al. generalized this by allowing arbitrary-size Mie or geometrical-optics particles described by unknown but linear scattering operators $$\Delta\phi_{\rm Sagnac}=\frac{4\pi nL}{\lambda}\theta_K,$$3 and $$\Delta\phi_{\rm Sagnac}=\frac{4\pi nL}{\lambda}\theta_K,$$4, by using orthogonally polarized CW and CCW beams to avoid standing-wave nodes and traps and to filter backscatter, and by using the parity operator $$\Delta\phi_{\rm Sagnac}=\frac{4\pi nL}{\lambda}\theta_K,$$5 to separate symmetric and antisymmetric components without requiring explicit field solutions. The approximation is small lateral displacement, $$\Delta\phi_{\rm Sagnac}=\frac{4\pi nL}{\lambda}\theta_K,$$6, and perfect mode overlap at the trap (Galinskiy et al., 2015).

In the realized back-focal-plane interferometry scheme, the photodetector is a First Sensor QP50-6 quadrant photodiode with bandwidth up to $$\Delta\phi_{\rm Sagnac}=\frac{4\pi nL}{\lambda}\theta_K,$$7 and saturation near $$\Delta\phi_{\rm Sagnac}=\frac{4\pi nL}{\lambda}\theta_K,$$8. The dark-port beam is attenuated to just below saturation, while the ghost-reflection beam is used for traditional BFP comparisons. The extracted observables are the “X-signal,” defined as left half minus right half photocurrent, and the “SUM,” defined as the total photocurrent. The normalized signal,

$$\Delta\phi_{\rm Sagnac}=\frac{4\pi nL}{\lambda}\theta_K,$$9

directly yields the SBR. Calibration is performed with a TTL-modulated side-pressure beam at $\theta_K$0, $\theta_K$1, and up to $\theta_K$2, which applies a known transverse oscillatory force at $\theta_K$3 with $\theta_K$4 duty cycle. The noise sources identified are optical shot noise, the electronic noise of the QPD amplifier, mechanical vibrations, and Brownian motion, with damping ratio $\theta_K$5 in air (Galinskiy et al., 2015).

The reported trap beam has $\theta_K$6 and total power $\theta_K$7 in the CW+CCW trap after the levitator split. It is focused by aspherical lenses of $\theta_K$8 with working distance $\theta_K$9. Particle delivery uses a commercial inkjet cartridge containing a $\phi_0(x)$0 glycerol solution, producing an initial droplet of $\phi_0(x)$1 diameter that is evaporatively reduced to $\phi_0(x)$2. The inkjet stream is tilted by $\phi_0(x)$3 into a vertical optical levitator of $\phi_0(x)$4 focal length and $\phi_0(x)$5 waist, enabling single-droplet capture on demand and stable trapping for minutes. For $\phi_0(x)$6 droplets, the trap stiffness is underdamped in air. The stated position resolution is sub-nanometer sensitivity limited by shot noise, with bandwidth up to $\phi_0(x)$7 set by the QPD and electronics. With a Thorlabs BSW26 beam splitter having $\phi_0(x)$8 and $\phi_0(x)$9, the theoretical maximum SBR gain is approximately $F[x_p]$0 and the SNR gain approximately $F[x_p]$1; experimentally, an SBR gain of $F[x_p]$2 and SNR improvement greater than $F[x_p]$3 were observed, consistent with theory (Galinskiy et al., 2015).

The proposal of Taylor et al. provides the asymptotic scaling of this gain. For a beam splitter of reflectivity $F[x_p]$4 and transmissivity $F[x_p]$5, the enhancement factor is

$F[x_p]$6

Introducing the interferometer visibility

$F[x_p]$7

one obtains

$F[x_p]$8

The maximum achievable enhancement is limited by detector saturation,

$F[x_p]$9

For $x_p$0, $x_p$1, and because Rayleigh scattering scales as $x_p$2, the minimum trackable particle radius is improved by

$x_p$3

This establishes the theoretical basis for the claim that a $x_p$4-visibility Sagnac-tweezers system can track particles about $x_p$5 smaller in radius than a conventional tweezers with identical trap power and detector (Taylor et al., 2010).

4. All-fiber Kerr microscopy

The all-fiber Sagnac Kerr microscope is a zero-area interferometer optimized for non-reciprocal magneto-optic phase. Active biasing is introduced by an electro-optic phase modulator driven at frequency $x_p$6 such that the two beams pass it half a period apart. Writing the modulation as $x_p$7, the detected intensity is

$x_p$8

which is expanded as

$x_p$9

Lock-in detection at $$F[x_p]\phi_0=\phi_s(x)+\phi_a(x), \qquad P\phi_s=+\phi_s,\quad P\phi_a=-\phi_a,$$0 isolates the term proportional to $$F[x_p]\phi_0=\phi_s(x)+\phi_a(x), \qquad P\phi_s=+\phi_s,\quad P\phi_a=-\phi_a,$$1, while detection at $$F[x_p]\phi_0=\phi_s(x)+\phi_a(x), \qquad P\phi_s=+\phi_s,\quad P\phi_a=-\phi_a,$$2 normalizes the total power. The standard readout is

$$F[x_p]\phi_0=\phi_s(x)+\phi_a(x), \qquad P\phi_s=+\phi_s,\quad P\phi_a=-\phi_a,$$3

The modulation depth is tuned to the first maximum of $$F[x_p]\phi_0=\phi_s(x)+\phi_a(x), \qquad P\phi_s=+\phi_s,\quad P\phi_a=-\phi_a,$$4, giving $$F[x_p]\phi_0=\phi_s(x)+\phi_a(x), \qquad P\phi_s=+\phi_s,\quad P\phi_a=-\phi_a,$$5, and the drive frequency is adjusted to $$F[x_p]\phi_0=\phi_s(x)+\phi_a(x), \qquad P\phi_s=+\phi_s,\quad P\phi_a=-\phi_a,$$6 so that the round-trip time is approximately $$F[x_p]\phi_0=\phi_s(x)+\phi_a(x), \qquad P\phi_s=+\phi_s,\quad P\phi_a=-\phi_a,$$7 (Fried et al., 2014).

The microscope is also a scanning instrument. A home-built “S-bender” piezo-bimorph scanner holds the fiber-lens/QWP assembly; two orthogonal bimorph pairs provide about $$F[x_p]\phi_0=\phi_s(x)+\phi_a(x), \qquad P\phi_s=+\phi_s,\quad P\phi_a=-\phi_a,$$8 stroke at room temperature, with somewhat reduced stroke at cryogenic temperature. Focus is provided by a vertical Attocube ANPxyz stage with travel $$F[x_p]\phi_0=\phi_s(x)+\phi_a(x), \qquad P\phi_s=+\phi_s,\quad P\phi_a=-\phi_a,$$9 and $1.5~\mu\mathrm{m}$00 resolution. The diffraction-limited spot size is estimated as

$1.5~\mu\mathrm{m}$01

so with $1.5~\mu\mathrm{m}$02 and $1.5~\mu\mathrm{m}$03, $1.5~\mu\mathrm{m}$04. In practice, lens aberrations and fiber mode mismatch yield a measured $1.5~\mu\mathrm{m}$05 FWHM by edge-scan of Au/GaAs (Fried et al., 2014).

The detection chain uses a GaAs photodiode with $1.5~\mu\mathrm{m}$06 bandwidth, responsivity $1.5~\mu\mathrm{m}$07, noise-equivalent power $1.5~\mu\mathrm{m}$08, and $1.5~\mu\mathrm{m}$09 load. Lock-in detection is performed at $1.5~\mu\mathrm{m}$10 and $1.5~\mu\mathrm{m}$11. A time constant of approximately $1.5~\mu\mathrm{m}$12 gives an equivalent noise bandwidth of about $1.5~\mu\mathrm{m}$13 and a pixel dwell time of approximately $1.5~\mu\mathrm{m}$14. Because the EOM insertion loss is about $1.5~\mu\mathrm{m}$15, the apparatus is detector-noise-limited, with typical $1.5~\mu\mathrm{m}$16. The reported performance is angular resolution $1.5~\mu\mathrm{m}$17 in $1.5~\mu\mathrm{m}$18, spatial resolution $1.5~\mu\mathrm{m}$19 FWHM, scan area up to $1.5~\mu\mathrm{m}$20 at room temperature, Kerr-signal drift less than $1.5~\mu\mathrm{m}$21 over $1.5~\mu\mathrm{m}$22, and reflectivity-contrast at $1.5~\mu\mathrm{m}$23 typically $1.5~\mu\mathrm{m}$24 from $1.5~\mu\mathrm{m}$25 incident, corresponding to $1.5~\mu\mathrm{m}$26 throughput (Fried et al., 2014).

5. Spin-Seebeck measurements and methodological controversy

A Sagnac interferometer microscope has also been used to probe the transverse spin-Seebeck effect in a Permalloy film with an in-plane temperature gradient. In the formulation summarized by Kimling and Kuschel, the interferometer measures a relative phase

$1.5~\mu\mathrm{m}$27

with detected intensity

$1.5~\mu\mathrm{m}$28

The magneto-optically induced phase contains the Kerr rotation, and in the presence of an in-plane temperature gradient $1.5~\mu\mathrm{m}$29 and an in-plane magnetization $1.5~\mu\mathrm{m}$30, a local spin accumulation gives an additional Kerr rotation $1.5~\mu\mathrm{m}$31, entering as $1.5~\mu\mathrm{m}$32. The conversion from interference-signal change to spin accumulation is written as

$1.5~\mu\mathrm{m}$33

so that

$1.5~\mu\mathrm{m}$34

McLaughlin et al. assume the same Kerr-sensitivity coefficient for equilibrium thermal demagnetization and for nonequilibrium TSSE magnetization changes (Kimling et al., 2018).

The quantitative dispute centers on the reported signal size. The measured Kerr-rotation change for a $1.5~\mu\mathrm{m}$35 in-plane gradient over a sample length of about $1.5~\mu\mathrm{m}$36 is summarized as $1.5~\mu\mathrm{m}$37. Under the above assumption, this implies $1.5~\mu\mathrm{m}$38. The comment notes that the Kerr-rotation change from uniform heating of the entire film by $1.5~\mu\mathrm{m}$39 is also about $1.5~\mu\mathrm{m}$40, while bulk Py has $1.5~\mu\mathrm{m}$41. It then compares this inferred nonequilibrium magnetization with the conventional spin-dependent Seebeck effect in a $1.5~\mu\mathrm{m}$42 Py film, written as

$1.5~\mu\mathrm{m}$43

with $1.5~\mu\mathrm{m}$44, $1.5~\mu\mathrm{m}$45, $1.5~\mu\mathrm{m}$46, and $1.5~\mu\mathrm{m}$47. To obtain $1.5~\mu\mathrm{m}$48 at the surface would require $1.5~\mu\mathrm{m}$49 across the $1.5~\mu\mathrm{m}$50 film. On that basis, Kimling and Kuschel conclude that the observed signal is “not or not only caused by the TSSE” (Kimling et al., 2018).

The controversy is methodological as much as quantitative. The comment states that no standard procedures or reference measurements are available for verification of such novel experimental approaches. It identifies the absence of an independent calibration of $1.5~\mu\mathrm{m}$51 versus $1.5~\mu\mathrm{m}$52, the possibility of artefacts from small out-of-plane temperature gradients, and prior TSSE searches in Pt/Py bilayers that saw no signal above $1.5~\mu\mathrm{m}$53 when out-of-plane gradients were controlled. The suggested checks are independent calibration of Kerr rotation per unit $1.5~\mu\mathrm{m}$54 using known spin currents or a reference structure, careful mapping and suppression of any vertical $1.5~\mu\mathrm{m}$55, and cross-checks with inverse spin Hall effect measurements under identical thermal conditions (Kimling et al., 2018).

6. Engineering constraints and adaptation pathways

The available literature repeatedly identifies mode matching, background suppression, and stability as the central engineering constraints. In the tweezers context, the experimental gain was reduced from the theoretical $1.5~\mu\mathrm{m}$56 to $1.5~\mu\mathrm{m}$57 by mode-matching imperfections between CW and CCW beams, lens aberrations at $1.5~\mu\mathrm{m}$58, and mechanical drifts in droplet delivery. In the earlier theoretical proposal, achieving $1.5~\mu\mathrm{m}$59 required excellent spatial- and polarization-mode matching, low optical loss, and mechanical stability, and even perfect visibility could not exceed $1.5~\mu\mathrm{m}$60. The same experimental summary also outlines how the counterpropagating-tweezers system could be adapted into a high-resolution Sagnac-interferometer microscope: replace the aspherical trapping lenses with a high-$1.5~\mu\mathrm{m}$61 microscope objective, maintain splitting and recombination at a non-polarizing beamsplitter, preserve polarization control with half-wave plates and a Brewster-tilted plate, and address several stated challenges—wavefront matching through a high-$1.5~\mu\mathrm{m}$62 objective, control of back-reflections, and interferometer stability against thermal and mechanical drifts, for which active feedback or common-path designs may help. For biological specimens, the cited challenges are depolarization and multiple scattering, with adaptive polarization control or index-matching media identified as possible responses. The same source lists further modifications: a fiber-coupled Sagnac loop for compactness, fast quadrant-camera detectors for two-dimensional mapping of lateral and axial motion, extension of the theory to axial displacement through Gouy-phase shifts in $1.5~\mu\mathrm{m}$63, and active stabilization of the dark port with piezo-mounted mirrors (Galinskiy et al., 2015, Taylor et al., 2010).

The all-fiber Kerr implementation provides a complementary set of practical prescriptions. Alignment proceeds by placing a mirror at the sample position, adjusting the QWP angle to maximize the $1.5~\mu\mathrm{m}$64 amplitude, fine-tuning the EOM drive frequency for the largest $1.5~\mu\mathrm{m}$65 signal, adjusting $1.5~\mu\mathrm{m}$66 on a known magnetic sample to maximize $1.5~\mu\mathrm{m}$67, setting the lock-in phases so that $1.5~\mu\mathrm{m}$68 is purely in phase, verifying that $1.5~\mu\mathrm{m}$69 on gold is $1.5~\mu\mathrm{m}$70, and validating the system with $1.5~\mu\mathrm{m}$71-cut quartz plus an Al mirror, which yields a measured Verdet of about $1.5~\mu\mathrm{m}$72 compared with $1.5~\mu\mathrm{m}$73 expected. The listed practical tips are to use $1.5~\mu\mathrm{m}$74 APC connectors to suppress back-scatter, keep the SLED and EOM in thermally stable enclosures, control spurious RF pickup with common-mode filters, verify that the Kerr sign flips when the QWP is rotated by $1.5~\mu\mathrm{m}$75, and, if offsets greater than $1.5~\mu\mathrm{m}$76 persist, modify the fiber layout to minimize differential strain birefringence (Fried et al., 2014).

Taken together, these implementations show that the microscope form of the Sagnac interferometer is not a single instrument class but a family of reciprocal-background-suppressing measurement systems. In optical tweezers, the key observable is the antisymmetric scattering field on a quadrant detector; in Kerr microscopy, it is the non-reciprocal phase associated with magneto-optic rotation; and in spin-caloritronic applications, the same interferometric sensitivity makes calibration and artefact rejection decisive. This suggests that the principal value of the Sagnac microscope lies in the controlled separation of weak signal channels from large reciprocal optical backgrounds, rather than in any single specimen type or readout modality.